CN110942271B - Intelligent transportation scheduling management method based on neural network genetic algorithm - Google Patents
Intelligent transportation scheduling management method based on neural network genetic algorithm Download PDFInfo
- Publication number
- CN110942271B CN110942271B CN201911265953.XA CN201911265953A CN110942271B CN 110942271 B CN110942271 B CN 110942271B CN 201911265953 A CN201911265953 A CN 201911265953A CN 110942271 B CN110942271 B CN 110942271B
- Authority
- CN
- China
- Prior art keywords
- population
- chromosome
- neural network
- genetic algorithm
- transportation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0835—Relationships between shipper or supplier and carriers
- G06Q10/08355—Routing methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Abstract
The invention discloses an intelligent transportation scheduling management method based on a neural network genetic algorithm, which comprises the following steps: a1, establishing a data module according to a transportation scheduling management system; a2, generating an initial population G; the starting population comprises N chromosomes; the chromosome encodes the material M (Ma and/or Mb) according to any length and according to the selected characteristics; constructing a neural network; the delivery point is a neural network input node, the receiving point is an output node, and the transfer station is a hidden layer; and A4, starting iteration. The neural network algorithm is added into the genetic algorithm, so that the defects that the calculation efficiency in the genetic algorithm is low, the genetic algorithm is easy to fall into local optimum, the convergence is difficult and the like are avoided, and premature convergence or a large amount of iterative recalculation is avoided. The fitness evaluation function evaluates from two opposite angles of economic applicable quantity e and time length t, 2 opposite functions are balanced with each other, and a fast and economic logistics scheduling scheme is realized.
Description
Technical Field
The invention relates to the technical field of conveying pipelines, in particular to a high-pressure hydrogenation large-caliber super-wall-thickness seamless tee pipe fitting.
Background
The genetic algorithm is based on the principle of high or low in self-science, and then is introduced into an optimization algorithm, genetic operations carried out in the evolution process comprise coding, selection, crossing, variation and survival selection of fitters, function derivation is not needed, and function continuity is not required, the phenomena of propagation, crossing, gene mutation and the like in the natural selection and natural genetic processes are simulated, a group of candidate solutions are reserved in each iteration, superior individuals are selected from solution groups according to a certain index, the individuals are combined by using genetic operators (selection, crossing and variation) to generate a new generation of candidate solution groups, the process is repeated until a certain convergence index is met, the fitness function is used as the basis, and the population group individual population is continuously optimized and gradually approaches to the optimal solution by carrying out the genetic operations on the individual populations. The genetic algorithm realizes the gene based on the codes, and then is integrated into a scheme to link the gene segments into chromosomes, and the fitness of each individual to the clustering problem is measured by constructing a fitness function, namely if the codes of a certain individual represent good clustering results, the fitness is high; otherwise, its fitness is low. The fitness function is similar to the action of the environment in the organism evolution process, individuals with high fitness generate more offspring in the reproduction process of one generation and the next generation, and individuals with low fitness gradually die; however, the genetic algorithm has the defects of low calculation efficiency, easy falling into local optimization, difficult convergence and the like, and can cause premature convergence or a large amount of iterative recalculation, in the modern logistics industry, the object flow is more and more large, the method is very consistent with the basic code of the genetic algorithm, the sorting is carried out from an unordered state, how to rapidly and effectively schedule vehicles and schedule vehicles as few as possible, the method is obviously a difficult problem measured among the distance, the delivery time and the economic suitability, and the method cannot be solved by the genetic algorithm.
Disclosure of Invention
The invention aims to provide an intelligent transportation scheduling management method based on a neural network genetic algorithm to solve the problems;
the method comprises the following steps:
a1, establishing a data module according to a transportation scheduling management system, and taking a transportation scheme of materials as a chromosome, wherein parameters comprise delivery point information, transfer station information, receiving point information, transportation tool information and material information as genes of the chromosome; the relevant symbols are as follows:
n: a delivery point comprising { n1, n2, n3 … … n }; the delivery point is used as an important gene of the material and is compiled into the gene, the delivery amount of a certain delivery point can be quickly searched, and the delivery amount of the certain delivery point is calculated; thus, the transportation means can be calculated and selected and can be programmed as genes;
m: a receiving point comprising { m1, m2, m3 … … m }; the receiving points are used as another important gene of the materials, the delivery points and the receiving points form a most basic chromosome (namely a material transportation scheme), the receiving amount of a certain receiving point can be quickly searched, the receiving amount of a certain receiving point is calculated, and accordingly a transport can be calculated and selected, and the transport is coded as the gene;
q: the transfer station comprises { q1, q2, q3 … … q }; in modern logistics, the transfer station plays a crucial role, and the materials of each delivery point are rearranged and sent to each receiving point according to the receiving points, so that the centralized receiving and sending functions of the materials are realized, and a large amount of repeated transportation is avoided;
r: a transport path; the transportation route is used as a mode for evaluating the transportation scheme, a delivery point is used as a starting point, a receiving point is used as an end point, and the transportation route passes through a transfer station or does not pass through the transfer station; the transport path has vectorial properties, that is, the transport is in a point-to-point direction, and for a stable logistics system, the delivery point n and the receiving point m are the same set, and a delivery point can be used as a receiving point of another delivery point, even a delivery point can be used as a receiving point of the delivery point.
M: material supply; the material comprises a material weight Ma; material volume Mb; when the goods and materials are dispatched, the weight and the volume of the goods and materials are limited from 2 dimensions, and the carrying capacity of the transport vehicle at one time is limited by the weight or the volume;
d: a vehicle; including { D1,D2,D3… … D }; the carrier is loaded into chromosome as a gene fragment, comprising 2 processes, carrier selection from delivery point to transfer station, carrier selection from transfer station to receiving station;
t: the time is long; the time length is taken as the sum of the single material scheduling transportation time and is taken as an important evaluation parameter for evaluating the actual efficiency of the freight scheme (chromosome);
e: the amount is economic and suitable; the economic applicable quantity is used as an evaluation value which adopts the least transportation means and the shortest transportation path for the unit total cargo quantity, and is used for evaluating R, D, t and evaluating a fitness function; the later iteration stability of the genetic algorithm is realized, and a large amount of iterations are avoided;
a2, generating an initial population G; the starting population comprises N chromosomes; the chromosome encodes the material M (Ma and/or Mb) according to any length and features selected as required to generate a material scheduling sequence; the chromosome comprises a plurality of genes, and the gene segments comprise one or more of n, m, q, R and D; and the following conditions are required to be satisfied:
f(D)≥Manxmyor f (D) ≧ Mbnxmy；x∈n,y∈m；
Initially encoding the weight or volume of the basic material of the chromosome; the same material can also be coded in 2 types based on weight and volume, so that the gene quantity of a population is enlarged, the same material is coded from 2 dimensions, and the material information can be more accurately expressed; the transportation tool can not exceed the rated carrying weight or the rated volume which can be carried by the transportation tool every time of transportation;
A3、constructing a neural network; the delivery point is a neural network input node, the receiving point is an output node, and the transfer station is a hidden layer; namely, the neural network has n input nodes, m output nodes and q neurons; the mapping relation of the chromosomes is { n }1,n2,n3……n}→{qxn1,qxn2,qxn3……qxn}→{qxqyn1,qxqyn2,qxqyn3……qxqyn}→{m1,m2,m3… … m }; wherein q is more than or equal to x is more than or equal to 1; q is more than or equal to y is more than or equal to 1; inputting and outputting the chromosome through a neural network to form a transportation scheme of the chromosome; an artificial neural network is a second way of simulating human thinking; the system is a nonlinear system and is characterized in that distributed storage and parallel cooperative processing of information are realized, although a single neuron has extremely simple structure and limited functions, the behavior realized by a network system formed by a large number of neurons simulates the normal behavior of logistics scheduling, and the neural network can adapt to the environment and summarize the rule and complete certain operation, identification or process control instead of executing operation step by step according to a given program. The neural network is trained through a large amount of data to obtain a mapping relation between input and output, the weight and the threshold value of the network are continuously adjusted through a gradient descent method to enable the error of the network to be minimum, the neural network algorithm is implanted into a genetic algorithm to form large-amplitude subtraction operation of the neural network algorithm on the genetic algorithm, the neural network is based on the gradient descent method, the convergence speed is low, and an error function is easy to fall into a local minimum value. The genetic algorithm searches from the population, does not find the optimal solution from one point, and therefore has good global optimization capability. The neural network is implanted into the genetic algorithm, so that the genetic algorithm is optimized, the operation data of the genetic algorithm is reduced, the iteration times are reduced, and the optimal solution can be obtained more quickly.
A4, starting iteration, and executing the following steps:
1) carrying out cross operation on chromosomes in the initial population according to the cross probability, and randomly exchanging gene segments in the cross process to form a population G1; the population G1 comprises a coding gene set based on the weight and volume of the material, and the sequence combination of n, m, q and D is carried out by taking Ma and Mb as the starting points of the gene segments; after crossing, the chromosome contains a plurality of genes in uncertain disorder states to form a plurality of scheduling schemes;
2) carrying out fitness evaluation on the population G1 according to the scheduling targets R, t and e to obtain a population G2; setting a scheduling target value and fitness evaluation functions of R, t and e, and screening genes in a population G1 to obtain an optimized population G2;
3) carrying out mutation operation on the population G2 according to the mutation probability; in order to avoid premature convergence or early stabilization of chromosomes in the population, the population G2 is subjected to mutation operation, so that the genes of the chromosomes are mutated, and the types of the chromosomes are enriched;
4) carrying out cross operation on chromosomes in the population according to the cross probability by using the population G2 mutated in the step 3), and randomly exchanging gene segments in the cross process to form a population G3; the gene mutation has uncertain factors, the mutation rate can be set, the mutation form is uncontrollable, the gene of the chromosome is crossed, the mutated gene is better stored, and the first screening after mutation is prevented from being eliminated by a fitness evaluation function;
5) carrying out fitness evaluation on the population G3 according to the scheduling targets R, t and e to obtain a population G4; evaluating the fitness of the mutated and cross-recombined chromosome again to reduce the calculation amount; thus, the first iterative operation of the chromosome is completed;
6) repeating the population iteration to obtain a population Gi, and judging whether a convergence optimization result is achieved; if yes, the iteration is ended, and if not, the iteration is continued. And (5) iterating the population for multiple times, evaluating the chromosome and the target value, and judging whether the expected value of convergence optimization is reached.
As a further improvement of the above scheme:
further, the method comprises the following steps of; the economic applicable amount is the unit total cargo amount and adopts the minimum transport means and the evaluation value with the shortest transport path;
Allocating the probability P that the economically applicable amount is selected according to the sequence:
the required values are matched according to the selected probability P, the P serves as a fitness function of the second level to schedule materials, the carrying capacity of the transport means (vehicles) is exerted to the maximum, the probability P reflects the closest matching value, therefore, the former P can meet the next selected P, the residual quantity is the minimum, namely the vehicle transport allowance is the minimum, and the carrying maximum is realized.
Further, the method comprises the following steps of; the chromosome is a threshold value in a neural network population, and the population of each iteration is a connection weight of the neural network;
G＝nq+qm+q+m。
the connection weight is a gene numerical value of the population, the connection weight of the gene is calculated, the weight or the threshold of the neural network is continuously adjusted, the error of the neural network is minimized, and the identification precision of the network is improved.
Further, the method comprises the following steps of; the chromosome of the connection weight is endowed to a neural network, and the square sum Q of an expected value y and an output value o of the chromosome is calculated;
and endowing the connection weight in the population to a neural network, and calculating the square sum of the output value of the ith receiving point and the error of the output value, wherein the smaller the square sum of the errors is, the smaller the fitness function value is, and the better the chromosome scheduling scheme is.
When the goods and materials sent by a certain delivery point can be delivered to a receiving point without passing through a transfer station, the smaller the Q value is, the more stable the Q value is, even when the delivery point and the receiving point of the goods and materials are the same station, the Q value is 0, namely, the transportation is not needed.
Further, the method comprises the following steps of; the chromosome variance is: s2Q/f; wherein f is a degree of freedom. And (3) subjecting the error sum of squares to variance, measuring the difference degree of the chromosomes, the degree of variance reaction and center deviation by using a standard deviation, and measuring the fluctuation size of the chromosomes (namely the deviation size of the data from the average), wherein under the condition that the sample connection weight is the same, the larger the variance is, the larger the fluctuation of the chromosomes is, the more unstable the chromosome is, and on the contrary, the smaller the variance is, the smaller the fluctuation of the chromosomes is, the more stable the chromosome is. f is used as a degree of freedom and can be adjusted according to expected values to match iteration of the genetic algorithm, so that premature convergence is avoided.
Further, the method comprises the following steps of; genes { n ] of chromosomes in the fitness evaluation step1,n2,n3… … n, i.e., no more than 3 delivery points in a gene at the most. Avoiding multiple consignments of a transport at a point of delivery, genes { n } in the chromosome1,n2,n3… … n, so that a certain transport vehicle can only go to 3 delivery points at most, thereby improving the receiving efficiency.
Further, the method comprises the following steps of; the minimum value of chromosome R is:
min(R)＝L1|nx-nx+1|+L2|nx+1-nx+2|+L3|nx+2-nx+3|+L4|nx+3-qy|+L5|qy-mzl, |; wherein L is1,L2,L3,L4,L5Is a constant value; i.e. the actual distance from one point to another;
wherein x belongs to n; y belongs to q; z ∈ m. Since the chromosome expresses a transportation scheme, the path from a certain delivery point to another delivery point is constant, the path from a certain delivery point to a certain transfer station is also constant, and the path from a certain transfer station to a certain receiving point is also constant, the minimum path value of the chromosome can be calculated to measure the minimum transportation path of the material.
Further, the method comprises the following steps of; the minimum value of the chromosome time length t is:
(t) min (r)/v- ρ Δ t; Δ t is a loading time of the delivery point, ρ is a number of times of loading, and ρ is 3 or less. Since the transport route of the chromosome is constant, the transport time can be calculated assuming that the transport speed is constant, and the gene expression information of the chromosome includes 1 delivery point or a plurality of delivery points, the number ρ of times of loading on the chromosome is also an exact value, and the time from the delivery point to the transfer station on the chromosome can be confirmed assuming that the loading time per delivery point is Δ t. The fitness function is used to evaluate shipping efficiency.
Further, the method comprises the following steps of; genes { m ] of chromosomes in the fitness evaluation step1,m2,m3… … m, no more than 3, i.e., no more than 3, delivery points in a gene at the most. Avoiding multiple deliveries of the vehicle at the point of receipt, genes { m } in the chromosome1,m2,m3… … m, so that a certain transport vehicle can only go to 3 receiving points at most, thereby improving the delivery efficiency.
Further, the method comprises the following steps of; the minimum value of chromosome R is:
min(R)＝L1|nj-qu|+L2|qu-mh|+L3|mh-mh+1|+L4|mh+1-mh+2|+L5|mh+2-mh+3l, |; wherein: l is1,L2L3,L4,L5Is a fixed value, i.e. the actual distance from one point to another;
wherein j belongs to n; u belongs to q; h is equal to m. Since the chromosome expresses a transportation scheme, the path from a certain delivery point to a certain transfer station is also constant, the path from a certain transfer station to a certain receiving point is also constant, and the path from a certain delivery point to another receiving point is also constant, the minimum path value of the chromosome can be calculated to measure the minimum transportation path of the material.
Further, the method comprises the following steps of; the minimum value of the chromosome time length t is:
f(t)＝min(R)/v-ρΔt1；Δt1ρ is the number of times of loading for the shipping point loading time, and ρ is 3 or less. Since the transport route of the chromosome is constant, the transport time can be calculated assuming a constant transport speed, and the gene expression information of the chromosome includes 1 receiving point or a plurality of receiving points, the number of times ρ of loading on the chromosome is also an exact value, and the loading time per delivery point is assumed to be Δ t1Therefore, the time from the delivery point to the transfer station in the chromosome can be confirmed. The fitness function is used to evaluate shipping efficiency.
Furthermore, the population after each iteration in the genetic algorithm can be added into the previous population for operation, so that premature convergence and inaccurate data are avoided. For example, the generated population obtained after the G1 operation is added into the G1 again to obtain a population G2.
Has the advantages that:
1. the neural network algorithm is added into the genetic algorithm, so that the defects that the calculation efficiency in the genetic algorithm is low, the genetic algorithm is easy to fall into local optimum, the convergence is difficult and the like are avoided, and premature convergence or a large amount of iterative recalculation is avoided.
2. Through the transportation scheme that a chromosome expresses a certain material, carry out gene coding with delivery point, transfer station, transport means and receiving point in the transportation for the chromosome has multiple gene, enriched the transportation scheme, the population of logistics scheduling has been enlarged, and simultaneously, delivery point, transfer station, and receiving point have constituted neural network's input, hidden layer and output, through adopting the mode of coding to encode the network connection weight, constantly adjust neural network's weight or threshold value, make neural network error reach the minimum, promote the identification precision of network. And performing error sum-of-squares and variance operations to quickly evaluate the degree of dominance and reaction of the scheduling scheme of the chromosome and the degree of center deviation and measure the fluctuation size of the chromosome.
3. The weight and the threshold value of the network are continuously adjusted through a gradient descent method, so that the network error is minimized, the neural network algorithm is implanted into the genetic algorithm, the large-amplitude subtraction operation of the neural network algorithm on the genetic algorithm is formed, the genetic algorithm is optimized, the operation data of the genetic algorithm is reduced, the iteration times are reduced, and the optimal solution can be obtained more quickly.
4. And evaluating the chromosome through a multi-layer fitness evaluation function, wherein a path function, a sorted and selected probability P function, a square sum Q function of an expected value y and an output value o of the chromosome, a variance function of the chromosome, a time length t, a minimum value function of the chromosome R and an economic suitability e function are used for evaluating the chromosome, and a scheme is rapidly selected from multiple dimensions.
5. The fitness evaluation function evaluates from two opposite angles of the economic applicable amount e and the time length t, the transportation time is long, 2 opposite functions are mutually balanced, and a fast and economic logistics scheduling scheme is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic structural view of the present invention.
FIG. 2 is a structural diagram of the fitness function of the present invention.
Fig. 3 is an iterative structure diagram of embodiment 1 of the present invention.
Fig. 4 is an iterative structure diagram of embodiment 2 of the present invention.
FIG. 5 is a schematic diagram of chromosome crossing operation of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the examples given herein without any inventive step, are within the scope of the present invention.
The technical solutions of the embodiments of the present invention can be combined, and the technical features of the embodiments can also be combined to form a new technical solution.
Example 1: as shown in fig. 1-3: the intelligent transportation scheduling management method based on the neural network genetic algorithm comprises the following steps:
a1, setting initialization parameters, establishing a data module according to a transportation scheduling management system, and taking a transportation scheme of materials as a chromosome, wherein the parameters comprise delivery point information, transfer station information, receiving point information, transportation tool information and material information as genes of the chromosome; the relevant symbols are as follows:
n: a delivery point comprising { n1, n2, n3 … … n };
m: a receiving point comprising { m1, m2, m3 … … m };
in the actual logistics scheduling, a delivery point and a receiving point are integrated, namely the delivery point is also used as the receiving point, the receiving point is also used as the delivery point, and n is equal to m; due to the material liquidity, 2 kinds of labels are provided for each site, namely a delivery point and a receiving point;
q: the transfer station comprises { q1, q2, q3 … … q }; the transfer station rearranges the materials of each delivery point according to the receiving points and sends the materials to each receiving point;
m: material supply; the material comprises a material weight Ma; material volume Mb; when the goods and materials are dispatched, the weight and the volume of the goods and materials are limited from 2 dimensions, and the carrying capacity of the transport vehicle at one time is limited by the weight or the volume;
d: a vehicle; including { D1,D2,D3… … D }; the transport vehicle has a direct relationship with the material, i.e., the transport vehicle has a nominal load bearing or nominal space;
r: a transport path; the transport route is one way to evaluate the transport solution.
t: the time is long; the time length is taken as the sum of the single material scheduling transportation time and is taken as an important evaluation parameter for evaluating the actual efficiency of the freight scheme (chromosome);
e: the amount is economic and suitable; the economic applicable quantity is used as an evaluation value which adopts the least transportation means and the shortest transportation path for the unit total cargo quantity, and is used for evaluating R, D, t and evaluating a fitness function; the later iteration stability of the genetic algorithm is realized, and a large amount of iterations are avoided;
a2, generating an initial population G; the starting population comprises N chromosomes; the chromosome encodes the material M (Ma and/or Mb) according to any length and features selected as required to generate a material scheduling sequence; the chromosome comprises a plurality of genes, and the gene segments comprise one or more of n, m, q, R and D; and the following conditions are required to be satisfied:
f(D)≥Manxmyor f (D) ≧ Mbnxmy(ii) a x belongs to n, y belongs to m; the chromosome has material information and information of a delivery point and a receiving point, namely, a complete transportation scheme is formed, wherein { n1, n2, n3 … … n } - { m1, m2, m3 … … m }, namely the delivery point and the receiving point can be the same site; also constitute a chromosome (material transportation scheme);
further, the method comprises the following steps of; chromosome is divided into basic gene ManxmyOr MbnxmyAnd also comprises one or more of q, R and D genes;
thus, within the chromosomeThe gene of (a) is the sum of { Ma, n, q, R, D, m } array sets, and the chromosome sequence is the permutation and combination of { Ma, n, q, R, D, m } array sets; such as: ma n2m3，Mb n5m7，Ma n2m3n1m8，Mb n1n4D3m7R9Wherein any one of n, q, R and D can appear for many times and then can be combined randomly; if n, a plurality of the genes can be contained in the same chromosome;
a3, constructing a neural network; the delivery point is a neural network input node, the receiving point is an output node, and the transfer station is a hidden layer; namely, the neural network has n input nodes, m output nodes and q neurons; the mapping relation of the chromosomes is { n }1,n2,n3……n}→{qxn1,qxn2,qxn3……qxn}→{qxqyn1,qxqyn2,qxqyn3……qxqyn}→{m1,m2,m3… … m }; wherein q is more than or equal to x is more than or equal to 1; q is more than or equal to y is more than or equal to 1; inputting and outputting the chromosome through a neural network to form a transportation scheme of the chromosome;
the neural network adapts to the environment by itself, summarizes the rules, and completes certain operations, identification or process control. The neural network is trained through a large amount of data to obtain a mapping relation between input and output, the weight and the threshold value of the network are continuously adjusted through a gradient descent method to enable the network error to be minimum, the neural network algorithm is implanted into the genetic algorithm, and the large-amplitude subtraction operation of the neural network algorithm on the genetic algorithm is formed.
A4, starting iteration, and executing the following steps:
1) carrying out cross operation on chromosomes in the initial population according to the cross probability, and randomly exchanging gene segments in the cross process to form a population G1; the population G1 comprises a coding gene set based on the weight and volume of the material, and the sequence combination of n, m, q and D is carried out by taking Ma and Mb as the starting points of the gene segments; after crossing (gene recombination, the random exchange recombination of genes in the chromosome), the chromosome contains a plurality of genes in uncertain disordered states to form a plurality of scheduling schemes;
2) carrying out fitness evaluation on the population G1 according to the scheduling targets R, t and e to obtain a population G2; setting a scheduling target value and fitness evaluation functions of R, t and e, and screening genes in a population G1 to obtain an optimized population G2;
3) carrying out mutation operation on the population G2 according to the mutation probability; in order to avoid premature convergence or early stabilization of chromosomes in the population, the population G2 is subjected to mutation operation, so that the genes of the chromosomes are mutated, and the types of the chromosomes are enriched;
4) carrying out cross operation on chromosomes in the population according to the cross probability by using the population G2 mutated in the step 3), and randomly exchanging gene segments in the cross process to form a population G3; the gene mutation has uncertain factors, the mutation rate can be set, the mutation form is uncontrollable, the gene of the chromosome is crossed, the mutated gene is better stored, and the first screening after mutation is prevented from being eliminated by a fitness evaluation function;
5) carrying out fitness evaluation on the population G3 according to the scheduling targets R, t and e to obtain a population G4; evaluating the fitness of the mutated and cross-recombined chromosome again to reduce the calculation amount; thus, the first iterative operation of the chromosome is completed;
6) repeating the population iteration to obtain a population Gi, and judging whether a convergence optimization result is achieved; if yes, the iteration is ended, and if not, the iteration is continued. And (5) iterating the population for multiple times, evaluating the chromosome and the target value, and judging whether the expected value of convergence optimization is reached.
Further, the method comprises the following steps of; the economic applicable amount is the unit total cargo amount and adopts the minimum transport means and the evaluation value with the shortest transport path;
Such as
If the value is 1.8, the lambda is 2, namely 2 vehicles are used for transporting materials, wherein 1 vehicle has the rest of 0.2 rated carrying weight or rated carrying space;
in this way, the goods and materials to be transported at each station can determine how many vehicles are needed, and the surplus of the goods and materials at each station after the transportation tool is matched can be determined;
the weight and volume of the material are used as basic information of the material, and can be recorded into the operation code when entering a delivery point.
The margin of each station is used as a carrying resource and cannot be wasted, so the probability P that the economic applicable quantity is selected is distributed according to the sequence:
In the actual logistics scheduling, perfect matching is few, so the probability P is used for matching the margin requirement relationship of the transport vehicles, the optimal solution is that 1 vehicle has a margin after being loaded at a certain delivery point, and the margin can also load materials from another delivery point nearby to the same transfer station.
The required values are matched according to the selected probability P, the P serves as a fitness function of the second level to schedule materials, the carrying capacity of the transport means (vehicles) is exerted to the maximum, the probability P reflects the closest matching value, therefore, the former P can meet the next selected P, the residual quantity is the minimum, namely the vehicle transport allowance is the minimum, and the carrying maximum is realized.
Further, the method comprises the following steps of; the chromosome is a threshold value in a neural network population, and the population of each iteration is a connection weight (gene) of the neural network;
G＝nq+qm+q+m。
the connection weight is a gene numerical value of the population, the connection weight of the gene is calculated, the weight or the threshold of the neural network is continuously adjusted, the error of the neural network is minimized, and the identification precision of the network is improved.
Further, the method comprises the following steps of; the chromosome of the connection weight is endowed to a neural network, and the square sum Q of an expected value y and an output value o of the chromosome is calculated;
the smaller the sum of squared errors of the chromosomes, the smaller the fitness function value, indicating the better the scheduling scheme of the chromosomes.
For example: desired value yiAnd the output value oiThe same, namely the material receiving point of the delivery point i is the delivery point, the Q value is 0, and the chromosome transportation scheme is optimal; the chromosome transport scheme also has a small Q value when the material is delivered to the point of delivery without passing through a transfer station.
Further, the method comprises the following steps of; the chromosome variance is: s2Q/f; wherein f is a degree of freedom. The sum of squared errors is squared, standard deviation is used to measure the degree of chromosome difference, the degree of variance response and center deviation is used to measure the fluctuation of chromosome (i.e. the deviation of the data from the average)Number size), the larger the variance, the more unstable the fluctuation of the chromosome, and conversely, the smaller the variance, the more stable the fluctuation of the chromosome. f is used as a degree of freedom and can be adjusted according to expected values to match iteration of the genetic algorithm, so that premature convergence is avoided.
Further, the method comprises the following steps of; genes { n ] of chromosomes in the fitness evaluation step1,n2,n3… … n, i.e., no more than 3 delivery points in a gene at the most. Avoiding multiple consignments of a transport at a point of delivery, genes { n } in the chromosome1,n2,n3… … n, so that a certain transport vehicle can only go to 3 delivery points at most, thereby improving the receiving efficiency.
Further, the method comprises the following steps of; the minimum value of chromosome R is:
min(R)＝L1|nx-nx+1|+L2|nx+1-nx+2|+L3|nx+2-nx+3|+L4|nx+3-qy|+L5|qy-mzl, |; wherein L is1,L2,L3,L4,L5Is a constant value; l is1|nx-nx+1I is nxTo nx+1The actual distance of (d); the positions of the screen dots are fixed, so the distance between the screen dots is also fixed.
Wherein x belongs to n; y belongs to q; z ∈ m. Since the chromosome expresses a transportation scheme, a path from a certain delivery point to another delivery point is constant, a path from a certain delivery point to a certain transfer station is also constant, and a path from a certain transfer station to a certain receiving point is also constant, the minimum path value of the chromosome can be calculated to evaluate the minimum transportation distance of the material.
When the delivery point is the receiving point, min (R) is also 0, when the materials do not need to pass through the transfer station, min (R) is that the delivery is realized among 2 stations or 3 stations, and the value of min (R) is the distance among the network points,
further, the method comprises the following steps of; the minimum value of the chromosome time length t is:
(t) min (r)/v- ρ Δ t; Δ t is a loading time of the delivery point, ρ is a number of times of loading, and ρ is 3 or less. Since the transport route of the chromosome is constant, the transport time can be calculated assuming that the transport speed is constant, and the gene expression information of the chromosome includes 1 delivery point or a plurality of delivery points, the number ρ of times of loading on the chromosome is also an exact value, and the time from the delivery point to the transfer station on the chromosome can be confirmed assuming that the loading time per delivery point is Δ t. The fitness function is used to evaluate shipping efficiency.
Further, the method comprises the following steps of; genes { m ] of chromosomes in the fitness evaluation step1,m2,m3… … m, no more than 3, i.e., no more than 3, delivery points in a gene at the most. Avoiding multiple deliveries of the vehicle at the point of receipt, genes { m } in the chromosome1,m2,m3… … m, so that a certain transport vehicle can only go to 3 receiving points at most, thereby improving the delivery efficiency.
Further, the method comprises the following steps of; the minimum value of chromosome R is:
min(R)＝L1|nj-qu|+L2|qu-mh|+L3|mh-mh+1|+L4|mh+1-mh+2|+L5|mh+2-mh+3l, |; wherein: l is1,L2L3,L4,L5Is a constant value, L4|mh+1-mh+2I is mh+1To mh+2The actual distance of (c).
Wherein j belongs to n; u belongs to q; h is equal to m. Since the chromosome expresses a transportation scheme, the path from a certain delivery point to a certain transfer station is also constant, the path from a certain transfer station to a certain receiving point is also constant, and the path from a certain delivery point to another receiving point is also constant, the minimum path value of the chromosome can be calculated to measure the minimum transportation path of the material.
Further, the method comprises the following steps of; the minimum value of the chromosome time length t is:
f(t)＝min(R)/v-ρΔt1；Δt1ρ is the number of times of loading for the shipping point loading time, and ρ is 3 or less. Since the transport route of the chromosome is constant, the transport time can be calculated assuming a constant transport speed, and the gene expression information of the chromosome includes 1 receiving point or a plurality of receiving points, the number of times ρ of loading on the chromosome is also an exact value, and the loading time per delivery point is assumed to be Δ t1Therefore, the time from the delivery point to the transfer station in the chromosome can be confirmed. The fitness function is used to evaluate shipping efficiency.
Example 2: the present embodiment is basically the same as embodiment 1, and the difference is that the population after each iteration in the genetic algorithm is added to the previous population for operation, so as to avoid premature convergence and cause data inaccuracy. For example, the generated population obtained after the G1 operation is added into the G1 again to obtain a population G2, and iteration is performed accordingly.
The invention is based on that when the goods and materials are sent to a delivery point, the goods and materials are weighed and subjected to volume accounting, and are coded according to the method, in the actual operation process, the goods and materials are scheduled to be t equal to 0h, 20 logistics network points in a certain area are selected for testing, and the testing is respectively carried out by using an embodiment 1, an embodiment 2, an independent genetic algorithm and an independent neural network algorithm:
| long average scheduling time | Long operation time | Required vehicle | Economic and applicable dosage | |
| Example 1 | 42.63h | 856s | 36 | 0.76 |
| Example 2 | 42.15h | 1235s | 35 | 0.73 |
| Genetic algorithm | 51.32h | 2864s | 39 | 0.97 |
| Neural network algorithm | 18.69h | 2563s | 38 | 0.95 |
As described above, the operation time of embodiment 1 is shorter than that of embodiment 2, and the scheduling time of embodiment 1 is longer than that of embodiment 2. And the method is obviously superior to a single genetic algorithm and a neural network algorithm, and has obvious advantages in solving a large number of logistics scheduling problems.
The technical solutions of the embodiments of the present invention can be combined, and the technical features of the embodiments can also be combined to form a new technical solution.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (3)
1. An intelligent transportation scheduling management method based on a neural network genetic algorithm is characterized by comprising the following steps:
a1, establishing a data module according to a transportation scheduling management system, wherein the parameters comprise delivery point information, transfer station information, receiving point information, transportation tool information and material information; the relevant symbols are as follows:
n: a delivery point comprising n1,n2,n3……n}；
m: a receiving point comprising { m1,m2,m3……m}；
q: a transfer station comprising { q1,q2,q3……q}；
R: a transport path;
m: material supply; the material comprises a material weight Ma; material volume Mb;
d: a vehicle; including { D1,D2,D3……D}；
t: the time is long;
e: the amount is economic and suitable;
a2, generating an initial population G; the starting population comprises N chromosomes; the chromosome encodes the material M according to any length and features selected as required to generate a material scheduling sequence; the chromosome comprises a plurality of genes, and the gene segments comprise one or more of n, m, q, R and D; and the following conditions are required to be satisfied:
f(D)≥Manxmyor f (D) ≧ Mbnxmy；x∈n,y∈m；
A3, constructing a neural network; the delivery point is a neural network input node, the receiving point is an output node, and the transfer station is a hidden layer; that is, the neural network has n input nodes, m output nodes, and q nervesElement; the mapping relation of the chromosomes is { n }1,n2,n3……n}→{qxn1,qxn2,qxn3……qxn}→{qxqyn1,qxqyn2,qxqyn3……qxqyn}→{m1,m2,m3… … m }; wherein q is more than or equal to x is more than or equal to 1; q is more than or equal to y is more than or equal to 1; inputting and outputting the chromosome through a neural network to form a transportation scheme of the chromosome;
a4, starting iteration, and executing the following steps:
1) carrying out cross operation on chromosomes in the initial population according to the cross probability, and randomly exchanging gene segments in the cross process to form a population G1;
2) carrying out fitness evaluation on the population G1 according to the scheduling targets R, t and e to obtain a population G2;
3) carrying out mutation operation on the population G2 according to the mutation probability;
4) carrying out cross operation on chromosomes in the population according to the cross probability by using the population G2 mutated in the step 3), and randomly exchanging gene segments in the cross process to form a population G3;
5) carrying out fitness evaluation on the population G3 according to the scheduling targets R, t and e to obtain a population G4;
6) repeating the population iteration to obtain a population Gi, and judging whether a convergence optimization result is achieved; if yes, finishing the iteration, and if not, continuing the iteration;
the economic applicable amount is the unit total cargo amount and adopts the minimum transport means and the evaluation value with the shortest transport path;
allocating the probability P that the economically applicable amount is selected according to the sequence:
the chromosome is a threshold value in a neural network population, and the population of each iteration is a connection weight of the neural network;
G＝nq+qm+q+m；
genes { n ] of chromosomes in the fitness evaluation step1,n2,n3… … n } no more than 3, i.e., no more than 3 delivery points in a gene at most;
the minimum value of chromosome R is:
min(R)＝L1|nx-nx+1|+L2|nx+1-nx+2|+L3|nx+2-nx+3|+L4|nx+3-qy|+L5|qy-mzl, |; wherein L is1,L2,L3,L4,L5Is a constant value; i.e. the actual distance from one point to another;
wherein x belongs to n; y belongs to q; z belongs to m;
the minimum value of the chromosome time length t is:
(t) min (r)/v- ρ Δ t; Δ t is a loading time of the delivery point, ρ is a number of times of loading, and ρ is 3 or less.
2. The intelligent transportation scheduling management method based on neural network genetic algorithm of claim 1, characterized in that: the chromosome of the connection weight is endowed to a neural network, and the square sum Q of an expected value y and an output value o of the chromosome is calculated;
the chromosome variance is therefore: s2Q/f; wherein f is a degree of freedom.
3. The intelligent transportation scheduling management method based on neural network genetic algorithm of claim 1, characterized in that; genes { m ] of chromosomes in the fitness evaluation step1,m2,m3… … m, no more than 3, i.e., no more than 3, delivery points in a gene at the most.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911265953.XA CN110942271B (en) | 2019-12-11 | 2019-12-11 | Intelligent transportation scheduling management method based on neural network genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911265953.XA CN110942271B (en) | 2019-12-11 | 2019-12-11 | Intelligent transportation scheduling management method based on neural network genetic algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110942271A CN110942271A (en) | 2020-03-31 |
| CN110942271B true CN110942271B (en) | 2021-04-09 |
Family
ID=69910128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911265953.XA Active CN110942271B (en) | 2019-12-11 | 2019-12-11 | Intelligent transportation scheduling management method based on neural network genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110942271B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111612124A (en) * | 2020-04-09 | 2020-09-01 | 杭州电子科技大学 | Network structure adaptive optimization method for task-oriented intelligent scheduling |
| CN114077911A (en) * | 2020-08-13 | 2022-02-22 | 福建金风科技有限公司 | Method and device for optimizing arrangement of transport paths of wind generating set |
| CN113222463B (en) * | 2021-05-31 | 2023-06-02 | 西安建筑科技大学 | Data-driven neural network agent-assisted strip mine unmanned truck scheduling method |
| CN115660538B (en) * | 2022-11-02 | 2023-07-21 | 广州和联慧通互联网科技有限公司 | Cargo transportation method and system |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105740961B (en) * | 2014-12-07 | 2021-08-06 | 中国石油化工股份有限公司 | Ship/vehicle-piecing transport optimization method for scattered goods based on evolutionary algorithm |
| CN105069518B (en) * | 2015-07-15 | 2018-10-30 | 西安交通大学 | A kind of distribution of materials method and system based on variable length Gene hepatitis B vaccine |
| CN106650981B (en) * | 2015-11-02 | 2020-08-11 | 中国石油化工股份有限公司 | Optimized purchasing and ship/vehicle splicing transportation method based on genetic algorithm |
| CN109800913B (en) * | 2019-01-12 | 2021-08-24 | 上海海事大学 | Intercommunication warehouse forklift operation line determining method for improving dangerous goods packaging in-out efficiency |
| CN110058573B (en) * | 2019-04-01 | 2022-04-15 | 青岛科技大学 | Huff and puff flexible intelligent assembly logistics path planning platform |
| CN110490503A (en) * | 2019-05-30 | 2019-11-22 | 湖南城市学院 | A kind of logistics delivery vehicle scheduling method based on mass data |
| CN110533279B (en) * | 2019-07-15 | 2023-01-06 | 合肥工业大学 | Method, system and storage medium for scheduling cloud logistics platform |
-
2019
- 2019-12-11 CN CN201911265953.XA patent/CN110942271B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110942271A (en) | 2020-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110942271B (en) | Intelligent transportation scheduling management method based on neural network genetic algorithm | |
| CN111027911B (en) | Automatic logistics storage transportation scheduling system | |
| CN111199375B (en) | Intelligent logistics transportation system | |
| CN111368999B (en) | Logistics scheduling system | |
| Khalili-Damghani et al. | A new bi-objective location-routing problem for distribution of perishable products: evolutionary computation approach | |
| Farahani et al. | A genetic algorithm to optimize the total cost and service level for just-in-time distribution in a supply chain | |
| CN107506846B (en) | Vehicle scheduling method and device based on multi-objective particle swarm optimization algorithm | |
| CN111311158B (en) | Electric logistics vehicle path planning method under limited charging facility condition | |
| CN111461402B (en) | Logistics scheduling optimization method and device, computer-readable storage medium and terminal | |
| CN109919532B (en) | Logistics node determining method and device | |
| CN103927628B (en) | Order management system and order management method that curstomer-oriented is promised to undertake | |
| CN110163544A (en) | A method of the genetic algorithm based on optimization carries out goods delivery | |
| CN113033895A (en) | Multi-source multi-point path planning method, equipment and storage medium | |
| CN110503225B (en) | Order dispatching and distributing method | |
| CN112183812A (en) | Finished cigarette logistics vehicle scheduling method considering short-time and low-cost | |
| CN113408771B (en) | Article delivery scheduling method, device and storage medium | |
| CN110942251B (en) | Batch scheduling method based on joint ant colony algorithm | |
| CN116644936A (en) | Path planning and scheduling method for multiple AGVs | |
| CN108108883A (en) | A kind of vehicle scheduling network resilience based on clustering algorithm simplifies method | |
| CN115330051B (en) | Vehicle path optimization method based on hybrid genetic algorithm and application | |
| Arab et al. | Solving a new multi-objective inventory-routing problem by a non-dominated sorting genetic algorithm | |
| Ashour et al. | Optimization algorithms for transportation problems with stochastic demand | |
| CN115952942A (en) | Multi-target inspection scheduling planning method and device, electronic equipment and storage medium | |
| Yu et al. | A hybrid genetic algorithm for integrated truck scheduling and product routing on the cross-docking system with multiple receiving and shipping docks | |
| CN115330288A (en) | Multi-supply-point emergency material site selection optimization method considering timeliness and fairness |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |